swansont

Prescription prices being high sounds like a US problem because our healthcare system sucks. In the civilized world they have national healthcare in various forms. But the problem of people trying quack cures because they don’t respond to real medical treatment is still there.

So? If the AI in question is to be installed on my computer, how does that matter?

In 10^10 years there will still be radioactive nuclei around. e.g. Rb-87’s half life is ~50 billion years. The list of isotopes with billion-year or longer half lives has quite a few members https://en.wikipedia.org/wiki/List_of_radioactive_nuclides_by_half-life

I think that’s just a matter of casual wording; one does not aim a bullet either, one aims the gun. But the OP also mentions pointing a light source at a target, which conveys the meaning, and something you can do. As far as “nothing fares from the laser to the screen” I disagree. You get an electric field in the intervening space, so a model of a light wave (or photons) traveling that path seems valid. Not sure where you’re going with this; photons most certainly carry linear momentum in addition to angular momentum. An atom absorbing or emitting a photon will recoil.

It’s offered as energy storage, so it’s pumped storage with a higher density than water. No 1st law issues.

My computer doesn’t work if it’s not plugged in, regardless of whether it’s hooked up to the internet.

You don’t legally get to wash your hands of responsibility that easily/casually, though. That’s one implication of the German court ruling. AI doesn’t turn the computer on and install itself.

I can see how you couldn’t measure time in that case, but is time dependent on that? Does time pass if there’s nothing around to experience it?

Stated without evidence, and contradicts observation Part of the issue with your previous posts was that you never adequately defined your terms. Precision is required, or the argument is susceptible to errors like the equivocation fallacy If you rewrote this mathematically, it would be a division-by-zero error; the ratio diverges. The implication of that is you can’t draw a valid conclusion. There are, of course, ways to have limits of such ratios converge, but it would be incumbent upon you to provide that rigor. The arguments presented are too sloppy.

Is there any credible discussion of time stopping in the heat death scenario?

As usual, though, this will come after damage has been done, even though some of the problems were easily foreseeable MEALworms are, by definition, too filling. The lembas bread of the larval insect world. You’d have to use snackworms.

I don’t understand the question. Who is “they” that’s writing code? Anthropic should be responsible for whatever its code does, and if someone uses AI to write code, that someone should be responsible for that code. Are you talking about writing code with AI and then blaming the AI for whatever bad happens as a result? Such lawsuits will better define who’s legally responsible for what, like how a manufacturer is liable for a defective product or certain obvious ways of using it unsafely, like with automobiles - there are legally required safety features, but the driver is at fault for accidents

Possibly related: a German court basically ruled that Google’s AI hallucinated results are statements by Google since they aren’t just information being passed along, so Google isn’t protected as normal search engine results are. “The court treated the AI overviews as Google's own content and rejected Google's argument that users were responsible for fact-checking the results themselves.” Also that allegedly being accurate 91% of the time isn’t good enough https://the-decoder.com/landmark-german-ruling-declares-googles-ai-overviews-are-googles-own-words-and-makes-it-liable-for-false-answers/ Since it ties responsibility for errors back to the creator of the AI, perhaps this is followed by similar rulings elsewhere and under other circumstances.

My equations are for conservation of momentum and energy; they already assume the actual device is as efficient as possible. The way to maximize efficiency is to have the rocket mass be small and the ejected mass be large. Such as a bomb set in between a payload and a really, really big rock. You drastically change the ratio - if you have a 1 kg payload and 9 kg rock, your efficiency is now 90%. But that also means a really large acceleration, and you have no more reaction mass (if you did, that has to be included with the payload mass so it’s not 90% efficient anymore), so no more thrust or maneuvering. Doesn’t work for the visitation scenario.

One of the issues with matter-antimatter is that it’s an energy storage medium, like a battery that needs charging, rather than tapping into already-stored energy. You need to create all the antimatter ahead of time, so that whole energy investment is up front. — Let’s look at a snapshot of a rocket that carries its own reaction mass. The ejected mass m is sent away at speed v, and the rest of the rocket has mass M and recoils at V. This is in the rest frame of the system as the mass is ejected. Conservation of momentum dictates mv = MV, so (mv)^2 = (MV)^2 Meaning the ratio of kinetic energy of each is KErocket/KEexhaust = m/M i.e. if, in some small time increment, you eject a kg of exhaust, your million kg of rocket (payload+fuel+superstructure, etc) gets one millionth of the energy of the exhaust. That’s a reason why rickets are multi-stage - you are shedding mass to improve the efficiency. But you can only do that if you don’t need that superstructure anymore, which is not the case if you want to, say, enter and leave some gravity well (or wells) after your trip. It only works as a one-off, like our trips to LEO, or to the moon, or probes/orbiters